From 92e83f064ade0f4a930722303a953326865600b2 Mon Sep 17 00:00:00 2001 From: Chris Gray Date: Sun, 15 Nov 2026 11:46:07 -0600 Subject: [PATCH 1/2] device/lynx: clarify historical Wi-Fi source and safety evidence Tie the two historical connection tests to an immutable archival source snapshot, preserve the DHCP-readiness failure and distinguish signed-source preparation from runtime qualification. Clarify the recorded Trusty-storage scope and hardware watchdog disarm, and credit alfikpl for the ordering lead. Assisted-by: LLM --- device/lynx/AOC-WIFI-PLAN.md | 97 +++++++++++++++++++++++++----------- device/lynx/NOTES.md | 24 +++++++++ device/lynx/WIFI-PLAN.md | 16 +++--- 3 files changed, 102 insertions(+), 35 deletions(-) diff --git a/device/lynx/AOC-WIFI-PLAN.md b/device/lynx/AOC-WIFI-PLAN.md index 8e789e9a2..93af7c479 100644 --- a/device/lynx/AOC-WIFI-PLAN.md +++ b/device/lynx/AOC-WIFI-PLAN.md @@ -1,24 +1,28 @@ # Pixel 7a: AoC before WLAN power -Status (2026-11-13): **one guarded association and traffic test passed on a -local experimental kernel**. This records a reproduction of the AoC-first -lead from [issue 36](https://github.com/derekross/omarchy-mobile/issues/36#issuecomment-6819528828) +Status (2026-11-15): **guarded association and traffic passed on two boots of +the old local experimental kernel**. This records our reproduction of alfikpl's AoC-first +ordering and four mainline fixes reported in [issue 36](https://github.com/derekross/omarchy-mobile/issues/36#issuecomment-6819528828) and the subsequent tests on the original lynx test phone. The kernel changes -and test overlays are not published or included here. No release package or -automatic driver startup is qualified by this result. +and test overlays for those historical runs are not included here. The revised +signed integration is now published as draft kernel PRs, but has never booted. +No release package or automatic driver startup is qualified by these results. -Recorded local build: detached kernel worktree -`~/android/mainline/wt-lynx-wifi-aoc-20261113`, with the base and local -changes described below; output `~/android/mainline/out-lynx-msi-20261113`. +The historical source and its distinction from the revised series are recorded +in section 1. Private build configuration and overlay receipts are retained +separately; no firmware or device capture is published. ## 1. Build and ordering -The successful connection ran on -`7.3.0-rc6-omamobile-lynx-msi-20261113+`, built from the published kernel -`omamobile` at `2db2bc35beecbf2deccb7f2e9bc8c3843b1edc9c` plus local -Trusty, AoC, power-sequencing, PCIe and board changes. That base commit alone -does not reproduce this build. The earlier firmware/scan build was +The two successful MSI-candidate connection tests ran on +`7.3.0-rc6-omamobile-lynx-msi-20261113+`. The exact historical kernel source is +[archival commit `3fd82de7aefe7eed42453271ef60ef6d29e4bbf0`](https://github.com/latentoperator/linux-omarchy-mobile/commit/3fd82de7aefe7eed42453271ef60ef6d29e4bbf0), +based on `2db2bc35beecbf2deccb7f2e9bc8c3843b1edc9c`. It preserves the complete +previously uncommitted tested tree, including Trusty, AoC, power-sequencing, +PCIe and board changes. This archival test branch is not proposed for merge +and is separate from the revised signed topic series. The base commit alone +does not reproduce the tested source. The earlier firmware/scan build was `7.3.0-rc6-omamobile-lynx-wifi-aoc-20261113+` on the same base with local changes, before the MSI edge-rearm candidate. @@ -42,6 +46,12 @@ data is redistributed. Bluetooth and the modem were not started for these tests. No extra chip-window, PMU, SysMMU or iATU userspace register reads were used to obtain the result. +The recorded lab stages did not start `storageproxyd` or any fingerprint or +modem service. Their preflights required those services inactive and rejected +a running storage proxy; the saved process checks found none. Trusty was used +for GSA/AoC startup. This describes the lab-stage scope, not an RPMB +write-counter measurement or a continuous whole-boot execution audit. + ## 2. Results, in order 1. **Firmware and scan passed.** With AoC online before WLAN power, Trusty @@ -68,25 +78,52 @@ were used to obtain the result. The final read-only checks found no failed units, all six required `phone-smoke` checks passing, the storage guard intact and all six protected -partition hashes matching the post-stock baseline. The watchdogs were -explicitly disabled after fresh healthy status; no stage guard or rollback -timer remained active. The successful connection stage did not reboot, +partition hashes matching the post-stock baseline. After fresh identity and +health checks, the successful stage explicitly disabled the S3C2410 hardware +watchdog it had opened as `/dev/watchdog0` with a 20-second timeout +(`WDIOS_DISABLECARD`), confirmed its disabled receipt and closed the descriptor. +Final checks found no watchdog descriptor owner and no active stage feeder +service or network rollback timer. This refers to that lab watchdog, not every +watchdog in the SoC. The successful connection stage did not reboot, reload firmware or modules, mark slot B, or write a protected partition. Optional smoke rows still reported a failed CMA reservation and no GPU render node; this Wi-Fi result does not qualify those paths. -This is **one successful runtime test** with the MSI edge-rearm change. -It does not establish why the earlier connection stalled or demonstrate -that the MSI change fixes it. The original failure and the successful -retest both remain part of the result. +4. **A later post-reboot restoration passed on the same old release.** Private + receipts saved on November 14 record a fresh guarded Trusty/AoC/link/firmware + sequence after the owner's reboot. AoC again reported 82 services and zero + restarts, and passive scan returned 21 BSS entries. The first activation + associated but its immediate IPv4 check raced DHCP completion and stopped; + Wi-Fi isolation rollback and watchdog disarm succeeded. The stage summary's + false association-request flag is stale: its primary per-boot marker and + kernel capture show that association was requested. + + A separately reviewed activation on that same boot added an eight-second + bound for IPv4 address, DHCP options and gateway readiness. It passed DHCP, + three gateway pings, Wi-Fi-bound HTTPS (HTTP 200, 577 bytes, TLS verified), + and fresh USB SSH. Ordinary HTTPS also used Wi-Fi and passed. The revised + activation did not reload modules or firmware or reboot. All six protected + hashes and required smoke checks passed, with no new WMI/TX or kernel error; + watchdogs and deadman timers were inactive at task close. Startup remained + manual after reboot; device autoconnect within that boot was enabled. + +These are **successful connection tests on two boots** with the MSI edge-rearm +candidate, including a preserved DHCP-readiness failure on the later boot. +They do not establish why the earlier WMI/TX connection stalled or demonstrate +that the MSI change fixes it. The original failure and successful retests all +remain part of the result. This is not a counted cold-start, reconnect or +endurance qualification series. ## 3. Current state and remaining gates The experimental drivers remain blacklisted for boot. Startup is a guarded -manual lab sequence. The owner subsequently rebooted the same kernel; -home and touch were confirmed working, and Wi-Fi startup has not yet been -repeated on that boot. The connection result above belongs to the earlier -qualified boot, not the phone's present network state. +manual lab sequence. The owner subsequently rebooted the same kernel; home +and touch were confirmed working, and the later restoration in section 2 +repeated guarded Wi-Fi startup on that boot. These are historical receipts, +not proof of present connectivity. The exact signed integration +`dfdf8ce251c89d3cd05a6325c61ed7ea1c61a741`, including the revised late-link/MSI +refusal paths and overlay split, has never booted and is not qualified by +either old-kernel result. Reconnect, repeated cold-start operation, sustained traffic, suspend/resume, Bluetooth coexistence and unattended startup remain unqualified. AoC @@ -103,8 +140,12 @@ On USB loss, stop sending commands and observe recovery before a fresh identity-pinned probe. Any boot-image write or reboot needs the owner's approval for that operation. -The next upstream code work is to separate the local kernel changes into -reviewable subsystem series, with explicit treatment of shared Trusty/AoC -paths and the Pixel 7 Pro. This documentation does not establish that those -paths are unchanged or ready to merge. Private logs, network identifiers, +The separated, human-certified topic series are published as draft kernel PRs +251-255. The signed integration now has a complete offline kernel/module +build and matched AoC-only/AoC+WLAN images, with image, AVB, module and overlay +verification. A newly reviewed attended runner and concrete owner approval +remain prerequisites for its first boot; offline preparation is not runtime +qualification. Shared Trusty/AoC paths still require a maintainer-approved +Pixel 7 Pro regression before merge. This does not establish that those paths +are unchanged or ready to merge. Private logs, network identifiers, partition images and calibration hashes remain off-repository. diff --git a/device/lynx/NOTES.md b/device/lynx/NOTES.md index 61f73744a..86da940d7 100644 --- a/device/lynx/NOTES.md +++ b/device/lynx/NOTES.md @@ -3,6 +3,30 @@ Findings from real hardware, newest first. Tracking issue: #36. +## 2026-11-14: later old-kernel Wi-Fi restoration receipts reconciled + +Private receipts saved after the earlier report record guarded startup after +the owner's reboot on `7.3.0-rc6-omamobile-lynx-msi-20261113+`: Trusty/AoC, +82 services with zero restarts, PCIe/MSI, one firmware load and passive scan +(21 BSS entries). An associated activation stopped when its immediate IPv4 +check raced DHCP; rollback/disarm succeeded. A reviewed activation with a +bounded DHCP-readiness wait then passed DHCP, gateway pings, Wi-Fi-bound and +ordinary HTTPS 200 with TLS verification, and fresh USB SSH. That revised +activation did not reload firmware/modules or reboot. All six protected +hashes and required smoke checks passed; no stage watchdog or network rollback +timer remained active at close. + +This adds a second boot's historical success and preserves its initial +failure. It does not qualify a counted cold-start/reconnect/endurance series, +prove current connectivity, or establish the MSI candidate as a stable fix. +Startup remained manual after reboot. The signed draft integration +`dfdf8ce251c89d3cd05a6325c61ed7ea1c61a741` has never booted; its revised +paths still need their own qualification. No hardware action was performed +to reconcile these receipts. Details and limits: [AOC-WIFI-PLAN.md](AOC-WIFI-PLAN.md). + +The November 13 entry below describes the earlier report's evidence cutoff; +its then-pending post-reboot startup is superseded by the later receipts above. + ## 2026-11-13: AoC-first Wi-Fi scan and one guarded traffic test - **AoC-first ordering reproduced on this lynx.** Trusty IPC and AoC came diff --git a/device/lynx/WIFI-PLAN.md b/device/lynx/WIFI-PLAN.md index 5a3862182..c2be178b2 100644 --- a/device/lynx/WIFI-PLAN.md +++ b/device/lynx/WIFI-PLAN.md @@ -1,15 +1,17 @@ # Pixel 7a (lynx): Wi-Fi and Bluetooth -Status (2026-11-13): **AoC-first firmware/scan reproduced; one guarded -association, DHCP and verified HTTPS test passed on a local experimental -kernel.** See [AOC-WIFI-PLAN.md](AOC-WIFI-PLAN.md) for the two kernel +Status (2026-11-14): **AoC-first firmware/scan reproduced; guarded association, +DHCP and verified HTTPS passed on two boots of the old experimental kernel.** +See [AOC-WIFI-PLAN.md](AOC-WIFI-PLAN.md) for the two kernel releases, the earlier WMI/TX stall, the successful retest and their limits. -The code and overlays are not published; boot-time startup, reconnect, +The revised signed integration is published as draft kernel PRs but has never +booted; boot-time startup, reconnect, sustained traffic and suspend/resume remain unqualified. The experimental -drivers stay blacklisted for boot. The owner has since rebooted; the guarded -startup has not been repeated on that boot, so this is a recorded test -result, not a claim that Wi-Fi is currently connected. +drivers stay blacklisted for boot. Later private restore receipts record +post-reboot guarded startup on the old release, including a DHCP-check race +and successful revised activation. These are recorded test results, not a +claim that Wi-Fi is currently connected or that the signed stack is qualified. Sections 1-8 below retain the investigation through 2026-11-05. Their parked state and next-step list describe that point in time; use the new AoC plan From f159a1b6dce9958385f34b0a6e3a08e7018e835f Mon Sep 17 00:00:00 2001 From: Chris Gray Date: Wed, 18 Nov 2026 20:38:15 -0600 Subject: [PATCH 2/2] device/lynx: document DMA limits of the historical Wi-Fi runs Record the endpoint and SysMMU configuration in the tested DT, date the draft kernel series and clarify the scope of the revised activation. Assisted-by: LLM --- device/lynx/AOC-WIFI-PLAN.md | 43 +++++++++++++++++++++++------------- device/lynx/NOTES.md | 8 +++---- device/lynx/WIFI-PLAN.md | 7 +++--- 3 files changed, 35 insertions(+), 23 deletions(-) diff --git a/device/lynx/AOC-WIFI-PLAN.md b/device/lynx/AOC-WIFI-PLAN.md index 93af7c479..7ff821123 100644 --- a/device/lynx/AOC-WIFI-PLAN.md +++ b/device/lynx/AOC-WIFI-PLAN.md @@ -5,8 +5,8 @@ Status (2026-11-15): **guarded association and traffic passed on two boots of the old local experimental kernel**. This records our reproduction of alfikpl's AoC-first ordering and four mainline fixes reported in [issue 36](https://github.com/derekross/omarchy-mobile/issues/36#issuecomment-6819528828) and the subsequent tests on the original lynx test phone. The kernel changes -and test overlays for those historical runs are not included here. The revised -signed integration is now published as draft kernel PRs, but has never booted. +and test overlays for those historical runs are not included here. The revised signed integration later +booted with AoC and WLAN held off (section 3). No release package or automatic driver startup is qualified by these results. The historical source and its distinction from the revised series are recorded @@ -30,8 +30,9 @@ The sequence was Trusty IPC, GSA/AoC, WLAN power, PCIe link, then ath11k. The local changes include: - Trusty IPC descriptors using premapped shared-memory buffer IDs; -- preserving GSA probe deferral in AoC and avoiding a second claim of the - chip-id resource already owned by exynos-chipid; +- preserving GSA probe deferral in AoC and mapping the chip-ID range without + requesting exclusive ownership (alfikpl reported an exynos-chipid conflict + on their build; this phone’s archived DT identifies no second owner); - an AoC-online dependency checked before WLAN power and PCIe bring-up; - lynx-specific WCN6855 power sequencing and DWC iMSI-RX handling, including restoring ELBI interrupt enables after link reset; @@ -44,7 +45,9 @@ firmware set: `amss20.bin`, `m3.bin`, `regdb.bin` and a `board-2.bin` packed from the stock board files for this endpoint. Neither firmware nor board data is redistributed. Bluetooth and the modem were not started for these tests. No extra chip-window, PMU, SysMMU or iATU userspace register reads -were used to obtain the result. +were used to obtain the result. In both MSI runs the WCN6855 endpoint had +no restricted DMA pool, and the PCIe controller had no active IOMMU mapping +(the HSI2 SysMMU was disabled in the tested DT). The recorded lab stages did not start `storageproxyd` or any fingerprint or modem service. Their preflights required those services inactive and rejected @@ -98,7 +101,7 @@ render node; this Wi-Fi result does not qualify those paths. false association-request flag is stale: its primary per-boot marker and kernel capture show that association was requested. - A separately reviewed activation on that same boot added an eight-second + A revised activation on that same boot added an eight-second bound for IPv4 address, DHCP options and gateway readiness. It passed DHCP, three gateway pings, Wi-Fi-bound HTTPS (HTTP 200, 577 bytes, TLS verified), and fresh USB SSH. Ordinary HTTPS also used Wi-Fi and passed. The revised @@ -120,10 +123,19 @@ The experimental drivers remain blacklisted for boot. Startup is a guarded manual lab sequence. The owner subsequently rebooted the same kernel; home and touch were confirmed working, and the later restoration in section 2 repeated guarded Wi-Fi startup on that boot. These are historical receipts, -not proof of present connectivity. The exact signed integration -`dfdf8ce251c89d3cd05a6325c61ed7ea1c61a741`, including the revised late-link/MSI -refusal paths and overlay split, has never booted and is not qualified by -either old-kernel result. +not proof of present connectivity. On November 15 the signed integration at +`dfdf8ce251c89d3cd05a6325c61ed7ea1c61a741` booted from B as +`7.3.0-rc6-omamobile-lynx-qual-dfdf8ce251c8+`. The first host check stopped +at an obsolete pinctrl debugfs path. A separate read-only check confirmed +the build and module set, with AoC and WLAN held off. Association and HTTPS +have only been demonstrated on the old experimental kernel. + +The board change still needs a restricted DMA pool that ath11k can use. +On this kernel MHI requests a contiguous 512 KiB firmware segment and the +restricted pool permits at most 256 KiB per allocation. Increasing the +total pool size would leave that limit in place. +[ath11k MHI configuration](https://github.com/latentoperator/linux-omarchy-mobile/blob/dfdf8ce251c89d3cd05a6325c61ed7ea1c61a741/drivers/net/wireless/ath/ath11k/mhi.c#L381-L384) +and [SWIOTLB segment and slot sizes](https://github.com/latentoperator/linux-omarchy-mobile/blob/dfdf8ce251c89d3cd05a6325c61ed7ea1c61a741/include/linux/swiotlb.h#L20-L32). Reconnect, repeated cold-start operation, sustained traffic, suspend/resume, Bluetooth coexistence and unattended startup remain unqualified. AoC @@ -140,12 +152,13 @@ On USB loss, stop sending commands and observe recovery before a fresh identity-pinned probe. Any boot-image write or reboot needs the owner's approval for that operation. -The separated, human-certified topic series are published as draft kernel PRs -251-255. The signed integration now has a complete offline kernel/module +As of 2026-11-15 the human-certified topics were open as drafts from +[linux-omarchy-mobile#251](https://github.com/derekross/linux-omarchy-mobile/pull/251) through +[linux-omarchy-mobile#255](https://github.com/derekross/linux-omarchy-mobile/pull/255). +That integration had a complete offline kernel/module build and matched AoC-only/AoC+WLAN images, with image, AVB, module and overlay -verification. A newly reviewed attended runner and concrete owner approval -remain prerequisites for its first boot; offline preparation is not runtime -qualification. Shared Trusty/AoC paths still require a maintainer-approved +verification. Testing AoC and Wi-Fi on that build still requires a revised +attended runner and owner approval. Shared Trusty/AoC paths still require a maintainer-approved Pixel 7 Pro regression before merge. This does not establish that those paths are unchanged or ready to merge. Private logs, network identifiers, partition images and calibration hashes remain off-repository. diff --git a/device/lynx/NOTES.md b/device/lynx/NOTES.md index 86da940d7..9d7bfee83 100644 --- a/device/lynx/NOTES.md +++ b/device/lynx/NOTES.md @@ -9,7 +9,7 @@ Private receipts saved after the earlier report record guarded startup after the owner's reboot on `7.3.0-rc6-omamobile-lynx-msi-20261113+`: Trusty/AoC, 82 services with zero restarts, PCIe/MSI, one firmware load and passive scan (21 BSS entries). An associated activation stopped when its immediate IPv4 -check raced DHCP; rollback/disarm succeeded. A reviewed activation with a +check raced DHCP; rollback/disarm succeeded. A revised activation with a bounded DHCP-readiness wait then passed DHCP, gateway pings, Wi-Fi-bound and ordinary HTTPS 200 with TLS verification, and fresh USB SSH. That revised activation did not reload firmware/modules or reboot. All six protected @@ -19,9 +19,9 @@ timer remained active at close. This adds a second boot's historical success and preserves its initial failure. It does not qualify a counted cold-start/reconnect/endurance series, prove current connectivity, or establish the MSI candidate as a stable fix. -Startup remained manual after reboot. The signed draft integration -`dfdf8ce251c89d3cd05a6325c61ed7ea1c61a741` has never booted; its revised -paths still need their own qualification. No hardware action was performed +Startup remained manual after reboot. The separate signed integration at +`dfdf8ce251c89d3cd05a6325c61ed7ea1c61a741` booted on November 15 with AoC +and WLAN held off. Its Wi-Fi paths still need their own qualification. No hardware action was performed to reconcile these receipts. Details and limits: [AOC-WIFI-PLAN.md](AOC-WIFI-PLAN.md). The November 13 entry below describes the earlier report's evidence cutoff; diff --git a/device/lynx/WIFI-PLAN.md b/device/lynx/WIFI-PLAN.md index c2be178b2..3cc737c17 100644 --- a/device/lynx/WIFI-PLAN.md +++ b/device/lynx/WIFI-PLAN.md @@ -1,13 +1,12 @@ # Pixel 7a (lynx): Wi-Fi and Bluetooth -Status (2026-11-14): **AoC-first firmware/scan reproduced; guarded association, +Status (2026-11-15): **AoC-first firmware/scan reproduced; guarded association, DHCP and verified HTTPS passed on two boots of the old experimental kernel.** See [AOC-WIFI-PLAN.md](AOC-WIFI-PLAN.md) for the two kernel releases, the earlier WMI/TX stall, the successful retest and their limits. -The revised signed integration is published as draft kernel PRs but has never -booted; boot-time startup, reconnect, -sustained traffic and suspend/resume remain unqualified. The experimental +The revised signed integration booted with AoC and WLAN held off. Its Wi-Fi +startup, reconnect, sustained traffic and suspend/resume remain unqualified. The experimental drivers stay blacklisted for boot. Later private restore receipts record post-reboot guarded startup on the old release, including a DHCP-check race and successful revised activation. These are recorded test results, not a